Letter

pubs.acs.org/ac

Think Negative: Finding the Best Electrospray Ionization/MS Mode

for Your Analyte
Piia Liigand,*,† Karl Kaupmees,† Kristjan Haav,† Jaanus Liigand,† Ivo Leito,† Marion Girod,‡
Rodolphe Antoine,¶ and Anneli Kruve†,§
†
Institute of Chemistry, University of Tartu, Ravila 14a, 50411, Tartu, Estonia
‡
Univ Lyon, CNRS, Université Claude Bernard Lyon 1, Ens de Lyon, Institut des Sciences Analytiques, UMR 5280, 5 rue de la Doua,
F-69100, Villeurbanne, France
¶
Univ Lyon, Université Claude Bernard Lyon 1, CNRS, Institut Lumière Matière, UMR 5306, F-69622, Lyon, France
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

§
Schulich Faculty of Chemistry, Technion - Israel Institute of Technology, Technion City, Haifa 3200008, Israel
*
S Supporting Information

ABSTRACT: For the first time, the electrospray ionization efficiency (IE) scales in
Downloaded via 181.209.93.106 on March 7, 2023 at 19:14:25 (UTC).

positive and negative mode are united into a single system enabling direct comparison
of IE values across ionization modes. This is made possible by the use of a reference
compound that ionizes to a similar extent in both positive and negative modes. Thus,
choosing the optimal (i.e., most sensitive) ionization conditions for a given set of
analytes is enabled. Ionization efficiencies of 33 compounds ionizing in both modes
demonstrate that, contrary to general practice, negative mode allows better sensitivity
for 46% of such compounds whereas the positive mode is preferred for only 18%, and
for 36%, the results for both modes are comparable.

M ass spectrometry (MS) with electrospray ionization

(ESI1) source is a key technique in various research
fields, ranging from food and environmental analysis to
metabolomics and proteomics.2 However, despite its widespread
application, current understanding of the ESI process is still
limited.3,4
Ionization efficiency (IE), the amount of ions generated from a
specific compound in the ionization source, may vary from
compound to compound by more than 6 orders of
magnitude.3,5−10 Further, IE in ESI is highly dependent on the
solvent3,5,11−14 and ionization mode.3 Higher IEs enable better
sensitivity and lower detection limits. Therefore, finding the
setup whereby the analyte of interest has the highest (or high
enough) IE allows lower limits of detections.
If ESI/MS is coupled with liquid chromatography (LC),
positive ion mode (ESI+) is generally preferred as more
compounds are expected to ionize in this mode.3,5 However,
the major advantage of negative ion mode (ESI−) is the lower
background noise.3,5 Currently, there is a paucity of research and
guidelines available5 on which mode to choose when a
compound ionizes in both modes.
In the interpretation of the ESI process, one important factor is
the solvent composition, which is usually described in terms of
initial composition, since the actual composition in the plume is
difficult to measure. However, it has been shown that solvent
pH,15−18 organic solvent content,19−23 and droplet size19,22,24,25
change along the plume. In a recent study,26 we have shown that
changes in pH, organic content, and droplet size are similar for
ESI+ and ESI−. This allows comparison of ESI+ and ESI−
modes without the need to account for major solvent property
changes between the two modes.
Absolute IEs are impractical to determine. Therefore, relative
ionization efficiencies (IE of the analyte relative to an anchor
substance) are commonly used. In both ESI+ and ESI−,
extensive IE scales have been constructed in 80/20 acetonitrile
(MeCN)/0.1% formic acid6 and 80/20 MeCN/0.1% ammonia
solution,7 respectively. All values on both scales are given relative
to another compound, for ESI+ relative to methyl benzoate
(MB) and in ESI− relative to benzoic acid (BA). However, the
ionization of methyl benzoate in ESI+ and benzoic acid in ESI−
cannot be compared directly.
In this study, we aimed for the first time to make ESI+6 and
ESI−7 IE scales quantitatively comparable in order to give further
insight into the differences and similarities of the ESI modes and
enable selection of the most suitable mode for any given
compound. This is accomplished (as depicted in Figure 1) by (1)
characterizing and comparing the ESI plume for ESI+ and ESI−,
(2) using a reference compound with both acidic and basic
moieties with similar pKa values, allowing ionization to a similar
extent in both modes from the same solution, and (3) measuring
Received: January 9, 2017
Accepted: May 10, 2017
Published: May 10, 2017

© 2017 American Chemical Society 5665 DOI: 10.1021/acs.analchem.7b00096

Anal. Chem. 2017, 89, 5665−5668
Analytical Chemistry
Figure 1. Process of unifying the different IE scales measured in different
solvents and in ESI+ and ESI−.
the links from ESI+ and ESI− anchoring compounds to the
reference compound (in respective solvents) and unifying ESI+
and ESI− scales.
■
EXPERIMENTAL SECTION
Ionization efficiencies were measured with an Agilent 6496
Triple Quadrupole mass spectrometer with Agilent Jet Stream
source (Agilent Technologies, Santa Clara, CA, USA) and
Agilent XCT ion trap spectrometer (see the Supporting
Information for further details). Relative logIE values were
obtained from the slope of the corresponding calibration graphs
of compound and anchor:
slope(compound)
log IE(compound) = log
slope(anchor)
where the anchor compound is BA (Reakhim, Russia) for ESI−,
in MeCN/0.1% ammonia solution 80/20, or tetraethylammo-
nium perchlorate (Fluka, Buchs, Switzerland) for ESI+, in
MeCN/0.1% HCOOH solution 80/20. The difference between
logIEESI− values in the two different solvents were obtained from
MS signal intensities for BA solution (concentration in both
solutions of approximately 60 μmol/L) according to eq 2 and
averaged over 3 days:
solvent(1) solvent(2)
solvent(1) RBA C BA
log IE BA = log solvent(2) solvent(1)
RBA C BA
where R is the response of the compound and C is concentration.
The degree of ionization of the reference compound, trans-
3(3-pyridyl)acrylic acid (TA) (Aldrich, St. Louis, USA), was
measured using 1H NMR and UV−vis spectroscopy (see the
details of the apparatus in the Supporting Information). For
NMR, the chemical shift of the protons was used to calculate the
degrees of ionization for the reference compound in the solvents
(based on the calibration with solutions with known degrees on
ionization; see Figure S1 and Table S2). A similar approach was
used for UV−vis.
In addition, 33 compounds (amino acids, substituted benzoic
acids and phenols, oligopeptides, and polyfunctional aromatic
compounds) were chosen for comparison of IE values in ESI+
and ESI−. See Table S1 for more details.
Weighted average positive and negative sigma (WAPS27 and
WANS,28 showing the degree of charge delocalization) values
were calculated with the COSMO-RS29 method. pKa values were
calculated with ACE and JChem acidity and basicity calculator30
(see the Supporting Information).
Letter
■ RESULTS AND DISCUSSION
Characterizing the ESI Plume. To make ESI+ and ESI−
scales comparable, it is important to find a system where logIEESI+
and logIEESI− would become similar (steps (2) and (3) in Figure
1). It has previously been observed7 that, among other things, the
ionization of compounds in the solvent and affinity toward the
droplet surface determine the extent of ionization in the ESI
source. Therefore, it was necessary that basic and acidic groups of
the anchoring compound would have similar ionization degrees
in solution. Since the physicochemical changes in the ESI plume
are similar in ESI+ and ESI−, the ionization in both modes
occurs from fairly similar solutions.26 This significantly simplifies
choosing a suitable solvent system for anchoring measurements.
Finding a Suitable Reference Compound. We tested
several compounds (see the Supporting Information) and
observed the closest degrees of ionization of the acidic and
basic groups for TA in 80/20 MeCN/pH 4.00 (v/v) solution.
The degree of ionization for the acidic group was 0.38 and for the
basic group, 0.19. Since the pH in the plume region close to the
MS entrance is somewhat lower (ca. 0.8 for ESI+ and 0.1 units for
ESI−26) than the initial pH, the degrees of ionization are in fact
even more similar and the ionization of TA in ESI+ and ESI− will
also be similar. Here, we focus on the ionization of a specific
group and not on the ionization of the compound as such, as
some of its molecules may be in solution as zwitterions. However,
the simultaneous ionization of two groups is expected to affect
ionization in both modes in the same way, as statistically the same
fraction of molecules is present in the droplets as zwitterions in
both modes.
Linking the Scales. The difference of anchor values
(ΔESI+ vs ESI−
anchors ) of ESI+ and ESI− can be derived (Figure 2):
(1)
ESI + vs ESI −
Δanchors = log IE EtESIN++,0.1%formic
− MB
+ log IE EtESIN++,0.1%formic − pH4
4 4
ESI+,pH4
+ log IE TA ESI−,pH4
− log IE TA ESI−,0.1%ammonia − pH4
− BA + log IE BA
− Et N+
4
(3)
(2)
Figure 2. Comparison of logIE scales compiled in ESI+ (logIEESI+) and
ESI− (logIEESI−).
where the superscript denotes the ESI mode and media that the
IE was measured in and the subscript denotes the compounds
that were measured.
TA−Et4N+ of TA was measured to be −1.11 in ESI+
The logIEESI+, pH 4
(anchor substance Et4N+) and logIEESI−, pH 4
TA−BA was 0.73 in ESI−
(anchor substance BA), both in 80/20 MeCN/pH 4.00 (v/v)
solution. However, logIE scales in ESI+ and ESI− have been
compiled in different solutions, and to make the scales
comparable, an anchor substance, BA in 80/20 MeCN/pH
4.00 (v/v) solution and 80/20 MeCN/0.1% ammonia (v/v)
solution (ESI−), was infused. A difference
ESI−, 0.1% ammonia−pH 4
(logIEBA ) of 0.90 logarithmic units was
observed, meaning that the IE of BA is 0.90 logarithmic units
5666 DOI: 10.1021/acs.analchem.7b00096
Anal. Chem. 2017, 89, 5665−5668
Analytical Chemistry
lower in 80/20 MeCN/pH 4.00. In ESI+, the Et4N+ in 80/20
MeCN/0.1% formic acid (v/v) solution has an IE of
(logIEESI+, 0.1% formic
Et4N+−MB ) 3.95 relative to methyl benzoate.6 It has
been observed previously that the IE of Et4N+ is not affected by
changes in pH (logIEESI+, 0.1% formic−pH 4
= 0).23
Et4N+
Unifying IE Scales. Substituting these values into eq 3 gives a
difference of 3.95 − 1.11 − 0.73 + 0.90 = 3.01 logarithmic units
between the IEs of anchors between ESI+ and ESI− (Figure 2).
Knowing the difference in anchors allows one to directly
compare logIE values between ESI+ and ESI− mode.
To further verify this difference, the logIE prediction models
were compared (eqs 4 and 5 in Figure 2). We have previously
developed a model (Figure 2, eq 5) for predicting IEs in ESI−7
(logIEESI−) based on the degree of ionization and the charge
delocalization in anions (WAPS).27 In order to make the IE
prediction models comparable, we have modified the model for
logIE values on the ESI+ scale6 to contain a WANS parameter
that describes charge delocalization in cations (see Table S3,
Figure S3)28 and obtained eq 4 (Figure 2) for finding logIEESI+.
The ability to predict IEs is similar (standard residual error, sRE =
0.87 logIE units) to the previous version6 of the model (sRE =
0.86 logIE units).
The intercept of such models carries information about the
logIE (relative to the anchoring compound) of a hypothetical
compound which is neutral (degree of ionization, α = 0) but has
infinitely delocalized charge (WANS or WAPS = 0). Such
compounds should ionize to the same extent in ESI+ and ESI−.
Therefore, the numerical value of the intercept carries
information about the anchoring compound, and the difference
in intercepts allows one to compare the anchoring values. The
difference in the intercepts (2.97) in the logIE prediction models
for ESI+ and ESI− (Figure 2) was statistically insignificantly
different (t-test on 95% confidence level) from Δanchors ESI+vs ESI−
(3.01). This implies that the difference is not accidental and
has been correctly assigned.
Comparison of the Two Ionization Modes. We applied
this knowledge to a set of 33 compounds (Table S1), which
ionize in both ESI+ (a range of 3.5 logIE units) and ESI− (a
range of 3 logIE units) and found the difference between
logIEESI+ and logIEESI− by taking into account the difference in
the anchors (Figure 3). The difference in logIE values (logIEESI+
Figure 3. Difference in logIE values for compounds ionizing in both ESI
+ and ESI− between the two modes.
− logIEESI−) was compared against a constant of 0.3 logarithmic
units, which refers to a two times difference in signal scale. We
find that a 2-fold increase is of significance for practitioners and is
also statistically significant (based on repeatability of the
measurements). Out of the investigated compounds, all of
which can be ionized in both modes, for six compounds, ESI+ is
preferred (logIE difference >0.3 log units), and for 15
compounds, ESI− is preferred (logIE difference < −0.3). Twelve
5667
Letter
compounds ionize to a similar degree in ESI+ and ESI−. Thus, in
approximately 46% of cases, a compound is better ionized in
ESI− and indifferent for 36%.
IE is considerably enhanced (up to 100 times) in ESI− for
compounds with low logIEESI+. More precisely, ESI− is preferred
by compounds that are only oxygen bases (e.g., mostly carboxylic
acids). Compounds being simultaneously both oxygen (carbox-
ylic acid) and nitrogen bases fall into all three categories. For
small peptides and amino acids, the differences between ESI+
and ESI− are mostly very small and either of the modes could be
used. One of the exceptions is histidine, which contains a basic
side chain that may account for a strong preference toward ESI+
mode (0.72 logIE units, 5.3 times). Aspartic acid and glutamic
acid, amino acids with acidic side chain, show weak preference
toward ESI−. When interpreting these results, it is of course
important to keep in mind that all the investigated compounds
have acidic groups in their structure.
From previous studies,6,7 we know that IE in both modes is
best described by degree of ionization in solution (α) and charge
delocalization (WAPS or WANS). For most compounds, the
ionization degree in solution is sufficient to explain the ionization
mode preference, including examples described above. For
example, compounds that are oxygen bases (such as carboxylic
acids) tend to be very weak bases but are at the same time
medium strength acids. Therefore, the formation of anions is
preferred for these compounds. For amino acids, both amino and
carboxylic acid group are expected to be charged, and therefore,
only small differences between ESI+ and ESI− are expected.
Exceptions to this rule are the amino acids tyrosine and
methionine. They are both fully ionized in ESI+ and ESI−
solvent; however, tyrosine is better ionized in ESI− and
methionine is better ionized in ESI+. To explain this
phenomenon, we need to look at the ionization process in
more detail. There are two main requirements for a compound to
become ionized: (1) it needs to become charged, and (2) it needs
to partition to the surface of ESI droplet to be ejected to the gas
phase. Previously, differences in anion and cation surface
affinities have been studied for protons and hydroxide ions31
and significant differences have been observed. These differences
are largely determined by ion−solvent interactions, which are
different for cations and anions of the same analyte, arising from
differences in charge delocalization, stereochemistry, and solvent
properties. In the case of methionine, the charge in the cation is
significantly better delocalized than in the anion, on the basis of
the comparison of WAPS and WANS (6.1 and 5.3, respectively; a
lower value means better charge delocalization). For tyrosine, the
charge is slightly better delocalized in the anion (6.0 and 6.3,
respectively). Therefore, the preferred ionization mode for
tyrosine is the opposite of methionine due to the reverse charge
delocalization of cation and anion. Additionally, tyrosine has an
ionizable side-chain in ESI−.32
■ CONCLUSIONS
ESI+ and ESI− logIE scales were made comparable, thereby for
the first time allowing comparison of ionization efficiency values
obtained in both modes. Until now, ESI+ has generally been
preferred among practitioners; however, it can be seen that, in
many cases, ESI− is the better option owing to its improved
sensitivity (ionization efficiency) and potential for lower
detection limits. While the solution phase ionization degree
mainly explains the ionization mode preference, solvent-induced
charge solvation of molecules in solution seems to be a key factor
driving the ionization efficiency of molecules in the ESI process.
DOI: 10.1021/acs.analchem.7b00096
Anal. Chem. 2017, 89, 5665−5668
Analytical Chemistry Letter

■
*
ASSOCIATED CONTENT
S Supporting Information
(15) Zhou, S.; Edwards, A. G.; Cook, K. D.; Van Berkel, G. J. Anal.
Chem. 1999, 71 (4), 769−776.
(16) Zhou, S.; Prebyl, B. S.; Cook, K. D. Anal. Chem. 2002, 74 (19),
The Supporting Information is available free of charge on the 4885−4888.
ACS Publications website at DOI: 10.1021/acs.anal- (17) Girod, M.; Dagany, X.; Antoine, R.; Dugourd, P. Int. J. Mass
chem.7b00096. Spectrom. 2011, 308 (1), 41−48.
Ionization efficiency measurements; physicochemical (18) Girod, M.; Antoine, R.; Dugourd, P.; Love, C.; Mordehai, A.;
Stafford, G. J. Am. Soc. Mass Spectrom. 2012, 23 (7), 1221−1231.
properties; determination of ionization degrees of trans- (19) Girod, M.; Dagany, X.; Boutou, V.; Broyer, M.; Antoine, R.;
3(3-pyridyl)acrylic acid; modified model for predicting Dugourd, P.; Mordehai, A.; Love, C.; Werlich, M.; Fjeldsted, J.; Stafford,
logIEESI+ values (PDF) G. Phys. Chem. Chem. Phys. 2012, 14 (26), 9389−9396.

■
(20) Zhou, S.; Cook, K. D. Anal. Chem. 2000, 72 (5), 963−969.
AUTHOR INFORMATION (21) Wang, R.; Zenobi, R. J. Am. Soc. Mass Spectrom. 2010, 21 (3),
378−385.
Corresponding Author (22) Hopkins, R. J.; Reid, J. P. J. Phys. Chem. A 2005, 109 (35), 7923−
*E-mail: piia.liigand@ut.ee. 7931.
ORCID (23) Liigand, J.; Kruve, A.; Leito, I.; Girod, M.; Antoine, R. J. Am. Soc.
Mass Spectrom. 2014, 25 (11), 1853−1861.
Piia Liigand: 0000-0001-7776-2650 (24) Wortmann, A.; Kistler-Momotova, A.; Zenobi, R.; Heine, M. C.;
Jaanus Liigand: 0000-0002-8814-9111 Wilhelm, O.; Pratsinis, S. E. J. Am. Soc. Mass Spectrom. 2007, 18 (3),
Ivo Leito: 0000-0002-3000-4964 385−393.
Rodolphe Antoine: 0000-0001-5682-8550 (25) Soleilhac, A.; Dagany, X.; Dugourd, P.; Girod, M.; Antoine, R.
Anal. Chem. 2015, 87 (16), 8210−8217.
Anneli Kruve: 0000-0001-9725-3351 (26) Liigand, P.; Heering (Suu), A.; Kaupmees, K.; Leito, I.; Girod, M.;
Author Contributions Antoine, R.; Kruve, A. J. Am. Soc. Mass Spectrom., submitted for
P.L. and K.K. performed UV−vis spectroscopy measurements. publication.
P.L., K.H., K.K., and A.K. performed NMR measurements. P.L., (27) Kaupmees, K.; Kaljurand, I.; Leito, I. J. Phys. Chem. A 2010, 114
J.L., and A.K. performed ionization efficiency measurements. P.L. (43), 11788−11793.
wrote the main part of the text, and K.K., I.L., M.G., R.A., and (28) Kaupmees, K.; Kaljurand, I.; Leito, I. J. Solution Chem. 2014, 43
(7), 1270−1281.
A.K. helped with preparing the manuscript. All authors have
(29) Klamt, A. COSMO-RS From Quantum Chemistry to Fluid Phase
given approval to the final version of the manuscript. Thermodynamics and Drug Design; Elsevier Science Ltd: Amsterdam,
Notes 2005.
The authors declare no competing financial interest. (30) ACE and JChem acidity and basicity calculator; https://epoch.

■
uky.edu/ace/public/pKa.jsp.
ACKNOWLEDGMENTS (31) Tse, Y.-L. S.; Chen, C.; Lindberg, G. E.; Kumar, R.; Voth, G. A. J.
Am. Chem. Soc. 2015, 137 (39), 12610−12616.
This work was supported by Personal Research Funding Project (32) Tian, Z.; Kass, S. R. J. Am. Chem. Soc. 2008, 130 (33), 10842−
34 from the Estonian Research Council. 10843.

■ REFERENCES
(1) Fenn, J. B. Angew. Chem., Int. Ed. 2003, 42 (33), 3871−3894.
(2) Doerr, A.; Finkelstein, J.; Jarchum, I.; Goodman, C.; Dekker, B.
Nat. Methods 2015, 12, S1−S45.
(3) Cech, N. B.; Enke, C. G. Mass Spectrom. Rev. 2001, 20 (6), 362−
387.
(4) Kostiainen, R.; Kauppila, T. J. J. Chromatogr. A 2009, 1216 (4),
685−699.
(5) Cole, R. B. Electrospray and MALDI Mass Spectrometry:
fundamentals, instrumentation, practicalities, and biological applications;
Wiley: Hoboken, 2010.
(6) Oss, M.; Kruve, A.; Herodes, K.; Leito, I. Anal. Chem. 2010, 82 (7),
2865−2872.
(7) Kruve, A.; Kaupmees, K.; Liigand, J.; Leito, I. Anal. Chem. 2014, 86
(10), 4822−4830.
(8) Raji, M. A.; Fryčaḱ , P.; Temiyasathit, C.; Kim, S. B.; Mavromaras,
G.; Ahn, J.-M.; Schug, K. A. Rapid Commun. Mass Spectrom. 2009, 23
(14), 2221−2232.
(9) Ehrmann, B. M.; Henriksen, T.; Cech, N. B. J. Am. Soc. Mass
Spectrom. 2008, 19 (5), 719−728.
(10) Henriksen, T.; Juhler, R. K.; Svensmark, B.; Cech, N. B. J. Am. Soc.
Mass Spectrom. 2005, 16 (4), 446−455.
(11) Zhou, S.; Hamburger, M. Rapid Commun. Mass Spectrom. 1995, 9
(15), 1516−1521.
(12) Zhou, S.; Cook, K. D. J. Am. Soc. Mass Spectrom. 2000, 11 (11),
961−966.
(13) Kruve, A. J. Mass Spectrom. 2016, 51 (8), 596−601.
(14) Huffman, B. A.; Poltash, M. L.; Hughey, C. A. Anal. Chem. 2012,
84 (22), 9942−9950.

5668 DOI: 10.1021/acs.analchem.7b00096

Anal. Chem. 2017, 89, 5665−5668